AI inference chip startup Etched raised $700 million [1] in a new financing round that pushed its valuation to $21 billion [2].
The rapid increase in valuation highlights a growing investor appetite for specialized AI hardware that can outperform general-purpose chips in efficiency and speed.
Quantitative trading firm Jane Street led the funding round. The investment followed a period where Jane Street tested the hardware and became the company's first paying customer [4, 5].
This latest capital injection comes shortly after a previous funding round in which Etched raised $300 million [1]. The company's valuation doubled to $21 billion [2] in under one month [3].
Etched focuses on the inference side of artificial intelligence, which is the process of using a trained model to make predictions or generate content. While many AI chips are designed for both training and inference, Etched develops hardware specifically for the latter to maximize performance.
The funding announcement was made Tuesday [2]. The company has not disclosed the specific technical specifications of the hardware that prompted Jane Street's investment, though the firm's transition from tester to paying customer served as a catalyst for the round [4].
“Etched raised $700 million in a new financing round that pushed its valuation to $21 billion.”
The surge in Etched's valuation suggests a shift in the AI hardware market toward 'domain-specific architectures.' By moving away from general-purpose GPUs toward chips optimized solely for inference, companies aim to reduce the massive energy and financial costs associated with running large-scale AI models. Jane Street's involvement is particularly notable because high-frequency trading firms require the lowest possible latency, indicating that Etched's hardware may offer significant speed advantages over current industry standards.
